TROLL model currently compute leaf lifespan with Reich’s allometry (Reich et al. 1991). But we have shown that Reich’s allometry is underestimating leaf lifespan for low LMA species. Moreover simulations estimated unrealistically low aboveground biomass for low LMA species. We assumed Reich’s allometry underestimation of leaf lifespan for low LMA species being the source of unrealistically low aboveground biomass inside TROLL simulations. We decided to find a better allometry with Wright et al. (2004) GLOPNET dataset.
Figure 1: Leaf mass per area (LMA), leaf nitrogen content (Nmass) and leaflifespan (LL). Leaf mass per area (LMA in \(g.m^{-2}\)), leaf nitrogen content (Nmass, in \(mg.g^-1\)) and leaf lifespan (LL in \(months\)) are taken in GLOPNET dataset from Wright et al. (2004).
We tested different models and we kept subsequant model with the best tradeoff between log likelihood, number of parameters, and convergence : \[ {LL_s}_j \sim \mathcal{logN}(log({\mu_s}_j),\,\sigma)\,\] \[s=1,...,S_{=4}~, ~~j=1,...,n_s, ~~X^{(1)}=LMA, ~~X^{(2)}=Nmass, ~~i=0,...,I_{=2}\]
\[{\mu_s}_j = {\beta_0}_s*e^{{\beta_1}_s*{X_s}_j^{(1)} - {\beta_2}_s*{X_s}_j^{(2)}}\] \[{\beta_i}_s \sim \mathcal{N}({\beta_i},\,\sigma_i)\,^I\] \[(\beta_i, \sigma, \sigma_i) \sim \mathcal{\Gamma}(0.001,\,0.001)\,^{2I+1}\] Leaf lifespan (\(LL\)) is following a lognormal law with location \(log({\mu_s}_j)\) and scale \(\sigma\). \(s\) represents the dataset used to do the fit between 1 and \(S=4\), it encompass environmental and protocol variations. \(j\) represents the observation for a given site \(s\). Finally \(X^{(1)}\) represents the LMA and \(X^{(2)}\) the leaf nitrogen content (Nmass). Location \({\mu_s}_j\) is the log of the exponential from the difference between LMA and Nmass with parameters \({\beta_0}_s\), \({\beta_1}_s\), and \({\beta_2}_s\). Each \({\beta_i}_s\) is following a normal law located on \(\beta_i\) with scale \(\sigma_i\). Finally all \(beta_i\), \(\sigma\), and \(\sigma_i\) are assumed without presemption following gamma laws of parameters \(10^-2\), \(10^-2\).
We obtained convergence for the model with a maximum likekihood of 10.0385366. Figure 2 shows posterior distributions and their associated markov chains. Figure 3 presents model predictions.
\[ LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA^{{\beta_3}_s} + {\beta_2}_s*N^{{\beta_4}_s},\sigma)\,\]
\[ LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA + {\beta_2}_s*N,\sigma)\,\] Maximum likekihood of 10.0385366
\[ LL \sim \mathcal{logN}(\beta_0 + LMA^{{\beta_3}_s} + N^{{\beta_4}_s},\sigma)\,\] Maximum likekihood of -1.731814
\[ LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA^{{\beta_3}_s} + {\beta_2}_s*N^{{\beta_4}_s},\sigma)\,\] Maximum likekihood of 18.3651541
\[ LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA + N,\sigma)\,\] Maximum likekihood of 0.880405
\[ LL \sim \mathcal{logN}(\beta_0 + LMA^{{\beta_3}_s} + N^{{\beta_4}_s},\sigma)\,\] Maximum likekihood of -0.1829464
\[ LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA^{{\beta_3}_s} + N^{{\beta_4}_s},\sigma)\,\] Maximum likekihood of 4.2540211
\[ LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA,\sigma)\,\] Maximum likekihood of 3.95695
\[ LL \sim \mathcal{logN}(\beta_0 + LMA^{{\beta_3}_s},\sigma)\,\] Maximum likekihood of -7.9790043
\[ LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA^{{\beta_3}_s},\sigma)\,\] Maximum likekihood of 9.4630726
Figure 2: Model posterior distributions and associated markov chains.
If we keep model 9: \(LL \sim \mathcal{logN}(\beta_0 + {\beta_1}_s*LMA^{{\beta_3}_s},\sigma)\,\)
Figure 3: Model predictions. \[LL = 1.529*e^{0.149 *LMA^{0.558}}\]
Reich, P.B., Uhl, C., Walters, M.B. & Ellsworth, D.S. (1991). Leaf lifespan as a determinant of leaf structure and function among 23 amazonian tree species. Oecologia, 86, 16–24.
Wright, I.J., Reich, P.B., Westoby, M., Ackerly, D.D., Baruch, Z., Bongers, F., Cavender-Bares, J., Chapin, T., Cornelissen, J.H.C., Diemer, M. & Others. (2004). The worldwide leaf economics spectrum. Nature, 428, 821–827.